AI Native Engineer, CookUnity

• Demonstrated ability to map a business process and redesign it with automation. This is the #1 skill. Show us a time you took a messy real-world workflow, understood it deeply, and made it dramatically better
• Technical fluency to connect systems and build production workflows. You're comfortable with APIs, CLIs, MCP, databases, and workflow orchestration. You can wire up business systems without waiting for someone else to do it
• Good judgment on agents vs. deterministic tools. You know when an LLM-powered agent is the right call and when a well-built internal app or script is better. You don't reach for AI when a cron job will do, and you don't build brittle rule engines when the task needs reasoning
• Strong foundation in at least one programming language (TypeScript, Python, Kotlin or similar) and ability to build full-stack when needed
• Hands-on experience with LLMs in production, not just prompting, but building reliable systems: structured outputs, eval frameworks, error handling, human-in-the-loop design
• Operational maturity. You don't just build things, you run them. You monitor, measure, iterate, and document
• Communication skills to work cross-functionally. You'll interview stakeholders, present business cases, and explain technical tradeoffs to non-technical leaders

Desirable

• You'll embed directly with the cross functional AI team as a dedicated AI engineer. Your job isn't to write code faster with AI
• It's to fundamentally redesign how work gets done. You'll map existing workflows end-to-end, identify where throwing compute at a task creates step-change value, and then build the solution: sometimes that's an agent, sometimes it's a deterministic internal app, sometimes it's a combination. Knowing which to reach for, and when, is a core part of the job
• You'll architect the automated version, build it, and then own it in production: tracking KPIs, managing evals after model changes, and continuously improving
• This is equal parts business analyst, systems integrator, and engineer. You'll spend as much time interviewing stakeholders and mapping data flows as you will writing code. You need to be the kind of person who can sit with a team lead, whiteboard a procurement process or a campaign launch workflow, spot the bottleneck that's begging for automation, and have a working prototype by end of week
• You'll report to the Head of Applied AI
• Map the team's highest-value workflows end-to-end: structured and unstructured data flows, decision points, handoffs, and bottlenecks
• Identify where automation can drive step-change outcomes: processing orders of magnitude more inputs, collapsing multi-day processes into hours, or enabling work that simply wasn't feasible before
• Design the future-state workflow: where an agent acts autonomously, where a deterministic app handles the logic, where humans review, what context is needed, and what guardrails keep it safe
• Build the business case. You can articulate why this workflow matters in terms of time saved, revenue impact, or cost reduced
• Build production workflows and internal tools that the team relies on daily, not just demos
• Use your judgment on the right tool for the job: agentic workflows when tasks require reasoning and flexibility, deterministic apps and scripts when the logic is well-defined and reliability matters most
• Work across the company's toolstack: APIs, CLIs, MCP servers, spreadsheets, data warehouses, workflow orchestration, databases, and whatever else needs connecting
• Create human-in-the-loop interfaces where they matter. Knowing when to automate fully and when to keep a human checkpoint is a core skill
• Write clean, maintainable code. You'll use AI tools to move fast, but you own the quality of what ships
• Own your agents and apps in production. Monitor performance, track KPIs, and explain outcomes to stakeholders
• Manage evals and regression testing after model updates, prompt changes, or upstream data shifts
• Continuously improve: every workflow you own should get measurably better over time
• Build institutional knowledge: document patterns, create playbooks, and teach your team to think in terms of what's now automatable
• Level up your team’s compute fluency and increase AI adoption
• What Success Looks Like
• Within 60 days, you've mapped 3 high value workflows and shipped your first production tool (agent or app) that the team uses daily
• Within 6 months, you've automated or agent-assisted 3-5 core workflows with measurable impact on team throughput or cost
• The team you've worked with considers you indispensable, not because you're the only one who can code, but because you see automation opportunities they can't
• You can show a clear before/after: hours saved, error rates reduced, throughput increased, or entirely new capabilities unlocked
• Your agents and apps are running reliably in production with monitoring, evals, and documented runbooks
• Technologies You'll Use
• LLM APIs: Anthropic (Claude), OpenAI, open-source models
• Agent and Orchestration: Claude Code, Codex App Server, Sandboxes, MCP, n8n, Temporal, Agent SDK, custom TypeScript/Python pipelines
• Data: Snowflake, Postgres, SQL, dbt
• Business Systems: Varies by team (Braze, Ramp, NetSuite, Jira, Linear, Slack, etc.)
• Collaboration: Slack, Notion, Google Workspace
• Infrastructure: Vercel, AWS